Prototipo de red neuronal convolucional artificial para la detección del cáncer de mama por medio del analisis de imágenes diagnosticas

Abstract

This study addresses a significant asymmetry in the education of medical professionals responsible for detecting breast cancer through imaging compared to the increasing number of new cases in Colombia. The resulting disproportionate workload for these doctors restricts access to timely diagnoses, leading to the loss of crucial benefits associated with early detection. Emphasizing the importance of improving the diagnostic process, this research aims to classify individuals as healthy or sick, enabling appropriate measures based on cancer stages. However, the slow pace of educating expert doctors in patient classification poses challenges in keeping up with the growing population requiring evaluation. Implementing a capable diagnostic system may extend access to early cancer detection benefits. The proposed application offers a tool for doctors to delegate tasks to less specialized medical personnel, increasing coverage. Nonetheless, it does not replace doctors' work, given unknown error rates. The project also aims to raise awareness of AI's potential in education and healthcare in Colombia. The upcoming prototype involves building a neural network from scratch using Python, including a relationship between models to train and test image classification.